The general goal of the proposed research is to determine how information about sensory stimulus features is encoded by ensemble neural activity patterns. The preparation to be studied is the cercal sensory system of the cricket. The specific aims of this proposal are to test the following hypotheses: 1) A dynamic temporal encoding scheme is used to represent features of sensory stimuli; 2) A collective encoding scheme is used to represent sensory stimuli; 3) There is no significant loss of behaviorally relevant information across the synaptic interface between the afferent array and the 20 cell ensemble of primary sensory interneurons; and, 4) The sensory system is optimized to process typical behaviorally relevant signals from the animal's environment. The experimental database for the studies will be collected through advanced electrophysiological recording techniques. Techniques will include dual intracellular recordings from pairs of neurons and multi-unit extracellular recordings from nerves. The extracellular recordings will be achieved through the use of wire and silicon recording arrays designed specifically for these studies. The analytical approach will employ information theory, stochastic systems analysis and several engineering analyses. Neural net algorithms may be used be implement nonlinear regression analyses of complex data sets.